Identification modeling of ship maneuvering motion based on physics-informed neural network
Description
This dataset contains the simulation data used in the study: “Identification modeling of ship maneuvering motion based on physics-informed neural network (PINN)”. The dataset includes simulated maneuvering motion data of the KVLCC2 ship used for training and validating the PINN model. The simulation data were generated using the Abkowitz-type ship maneuvering model with hydrodynamic derivatives obtained from literature. The dataset contains the state variables and control inputs used for system identification, including surge velocity (u), sway velocity (v), yaw rate (r), rudder angle (δ), and the corresponding accelerations. The simulation data include three maneuvering conditions: • 35°turning circle maneuver • 10°/10°zigzag maneuver • 20°/20°zigzag maneuver Each maneuver was simulated for 1000 s with a sampling interval of 1 s, resulting in a total of 3000 samples used for training. The experimental validation data used in the paper are obtained from the SIMMAN 2008 workshop database (http://www.simman2008.dk/) and are not included in this repository. Interested readers should obtain these data directly from the official SIMMAN database. These data are provided to support the reproducibility of the numerical experiments reported in the paper.
Files
Steps to reproduce
See the associated publication for the procedure used to reproduce the results.